The role of soil water in stormflow generation in a forested headwater catchment: synthesis of natural tracer and hydrometric evidence

Quantifying the role of soil water in stormflow generation is a difficult but important goal for addressing catchment scale environmental problems. The purpose of this study was to combine natural tracer methods and hydrometric observations to estimate the contribution of pre-event (vadose) soil water to stormflow in a small forested headwater catchment in Shenandoah National Park, Virginia. Two storm hydrographs were separated, based on mass balance of naturally occurring 18O and chloride, into three physically based source components. For a large storm event in June 1992, pre-event soil water contributed 36% of total storm runoff and dominated peak flow (65%), based on the hydrograph separation. For a smaller storm event in November 1992, soil water contributed 25% of the total runoff and 50% during peak flow. A Monte Carlo error analysis of the hydrograph separations supported the interpretation for the June storm that pre-event soil water dominated peak flow (alpha = 0.12), but showed considerable uncertainty--the estimated soil water contribution ranged from 48% to 88% at peak flow at a confidence level of 68%--even under the relatively good conditions (i.e. relatively distinct source component chemical and isotopic compositions). Analysis of the November storm indicated greater uncertainty and a more poorly identifiable soil water component. The June storm was typical of 'sharp' responses that occurred coincident with the development of a transient saturated zone above the soil-bedrock interface on hillslopes, whereas the November storm was representative of slower delayed responses in which the hydrograph peak occurred 8-10 h after the end of precipitation. During the 6 month study, five other hydrographs displayed the 'sharp' response and three other hydrographs showed the delayed response. During all storms, water draining from the soil into zero-tension lysimeters maintained the isotopic content of pre-event soil water; this result was taken as evidence that transient saturated zones would consist of nearly all pre-event soil water. Field observations after large storm events suggest a threshold type expansion of the source area onto the hillslopes which may supplement rates of downslope flow of displaced pre-event soil water. This study provides evidence that mobilization of pre-event soil water may dominate peak stormflow in steep forested headwater catchments, and illustrates that quantitative error analyses are advisable for hydrograph separation studies.